Ayoub et al hypothesize that one reason for the high observed rate of flares in our study is the possibility that 1 still-evolving flare could have been counted as 2 or more flares in our analysis if it occurred over several months. This is an interesting hypothesis, which we have assessed with our data. We found that, of the 65 flares defined by an increase of Ն1.0 in the physician's global assessment, there was only 1 instance of 2 flares occurring during consecutive months. Of the 108 flares defined by an increase in the SLE Disease Activity Index (Bombardier C, Gladman DD, Urowitz MB, Caron D, Chang CH, and the Committee on Prognosis Studies in SLE. Derivation of the SLEDAI: a disease activity index for lupus patients. Arthritis Rheum 1992;35:630-40), 7 occurred during the month following a previous flare. Thus, even if these should be considered as single flares rather than 2 or more, this does not completely explain the high observed flare rate.Another possible explanation for the high observed rate of flares is the fact that we measured disease activity monthly, rather than quarterly as was done in previous work cited by Ayoub and colleagues (Petri M, Genovese M, Engle E, Hochberg M. Definition, incidence, and clinical description of flare in systemic lupus erythematosus: a prospective cohort study. Arthritis Rheum 1991;34:937-44). Since we defined flare as an increase in disease activity over the previous measurement, there is obviously the potential to observe more flares with monthly measurements than with quarterly measurements. We believe that this change in methodology is the primary explanation for our high observed rates.A final possible explanation is that the flare rate may have been higher because we oversampled patients with abnormal anti-dsDNA or low complement levels. In our study, those who never had anti-dsDNA demonstrated by Crithidia assay had a flare rate (by physician's global assessment) of 1.0 per year, versus 1.7 per year if they had been positive for anti-dsDNA. Similarly, comparing the flare rates in patients with no anti-dsDNA by enzyme-linked immunosorbent assay versus those with a positive result, the respective flare rates were 1.3 per year and 1.6 per year.A flare is a difficult event to define, and it is clear that the rate of flare depends greatly on the definition and the frequency of measurements. Fortunately for us, the goal of this research was not to estimate absolute flare rates, but to explore the relationship between predictors and the occurrence of flare. We believe these relationships would be relatively invariant with different definitions of flare. We found similar relationships, for example, when we changed the cutpoints used to define flare. Thus, we do not believe the concerns raised by Ayoub et al should greatly affect the interpretation of our results.